The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes local CLI commands including obsidian-wiki, grep, and python3. These tools are used to query a session graph, verify file integrity, and parse JSONL transcript files to extract message content.
[PROMPT_INJECTION]: The skill has an indirect prompt injection surface because it ingests and processes past session transcripts, which could contain malicious instructions from previous untrusted interactions.
Ingestion points: Transcript files are read via obsidian-wiki sessions-show and then parsed by a Python script in Step 4.
Boundary markers: While the skill instructs the agent to prefix loaded content with > and apply a 400-character limit, it does not provide explicit instructions to the agent to disregard any embedded commands found within the historical transcript data.
Capability inventory: The skill uses obsidian-wiki for session retrieval and python3 for text extraction. It is intended to be read-only.
Sanitization: The Python script filters for user messages and basic metadata tags, but it does not perform sanitization or escaping of the message content to prevent potential injection patterns from being re-activated in the current context.

session-search